Embed Size (px)
Citation preview
Recent advances in development and application of assays/algotithms for
detection of recent HIV infections and estimation of incidence
Michael Busch, MD, PhDBlood Systems Research Institute
University of California San Francisco
CDC/APHL HIV Diagnostics Conference, 2010
Why its important to detect pre-SC window-phase HIV infections?
• Prevent viral transmission by blood transfusions and organ and tissue transplants
• Identify early infections in public health screening and diagnostic settings – acute-viremic phase of infection is highly
infectious– more effective response to treatment if initiated
during acute compared to chronic stages? – prevent secondary transmission by contact tracing
and counseling to modify risk behaviors• Identify subjects in primary infection for
pathogenesis, treatment and vaccine research
HIV acute and early infection
HIV RNA (plasma)HIV Antibody
11
0 10 20 30 40 50 60 70 80 90 100
HIV p24 Ag
16 22
Ramp-up viremiaDT = 21.5 hrs
1st gen2nd gen
3rd gen
p24 Ag EIA -
HIV MP-NAT -
HIV ID-NAT -
Peak viremia: 106-108 gEq/mL
Viral set-point: 102 -105 gEq/mL
Closing the infectious window period
HIV-1 transmission by transfusion of blood from SC donors according to the interdonation interval
Interdonation interval (days) Total
HIV-1 transmission
Number Percentage
45 - 90 17 13 76
91 - 180 29 8 28
181 - 360 48 9 19
361 - 540 39 5 13
541 - 720 14 0 0
> 720 32 1 3
Total 179 36 20
Peteresen et al. Transf. 1994 34 : 283-9Interdonation interval (days)
Prob
abilit
y of
tra
nsm
issi
on
0.0
0.2
0.4
0.8
1.0
0.6
2000 500400300100
56 day infectious WP (42 days w/ EIA 2.0)
HIV Stage Progression based on 51 Seroconverting Plasma Donors
Fiebig et al. AIDS, 17:1871-9, 2003Lee et al. J Theor Biol, 2009
The Incidence Rate / Window Period (WP) Model Allows Prediction of Test Yields for Direct HIV
Assays (p24 Ag, HIV RNA) vs. EIA Antibody Test
Test Yield (per unit) =
Incidence Rate (person-years) x Decrease in WP (fraction of year)
Projected WP Closure and Yield of p24 Ag, MP and ID NAT Assays Relative to a Sensitive HIV-1/2 EIA Antibody Test in the Detection of WP HIV Infection
Assay Sensitivity
[gEq / mL]
WP Closure
[days]
Yield, WP HIV Infections per 1,000 Persons Tested in Various Screening Settings
[ Representative Incidence Rate / Person-Years ]
Blood Donors[ 2 / 100,000 = 0.002% ]
STD Clinic[ 1 / 1,000 = 0.1% ]
High Risk Clinic[ 1 / 10 = 10% ]
p24 Ag 10,000 6 0.00033 0.016 1.6
MP NAT 1,000 9 0.00049 0.025 2.5
ID NAT 50 13 0.00071 0.036 3.6
Fiebig et al. AIDS, 17:1871-9, 2003
102 acutely infected plasma donor panels
3476 complete env sequences from single genome amplifications
Inferred consensus sequence at estimated time of virus transmission
78 donors infected by single virion; 24 by 2-5 virions
Why Determine HIV Incidence?• Characterize the epidemic in a population
– Monitor changes over time– Identify important sub-populations for interventions
• Assess impact of programs• Identify populations for HIV intervention trials
– Endpoint of intervention trials• Identify individuals for interventions
– Prioritization– Interrupt transmission
Standard Methods for Incidence Determination are Unsatisfactory
• Indirect methods; repeat cross-sectional measurements; modeling
• Prospective follow-up is expensive and unrepresentative
• Enrollment in cohorts leads to behavior change• Back calculation methods not timely or
reliable
HIV Incidence Using Early Diagnostic Tests
RNA+ Ab-P24+ Ab -
RNA
p24
Ab
seroconversion
days
Res
pons
e
infection
Detection limit
Brookmeyer, Quinn.Am J Epi 1995
Recent Infection Testing Algorithm (RITA)
RITA duration
RNA
p24
Ab
seroconversion
Antibody cutoff:Quantity (LS-EIA)Proportion (BED)AvidityIsotypeSpecificity of Ag
days
Res
pons
e
infection
Cross-Sectional Incidence Formula
AnnualizedIncidence =
(# who test recent) X (365/window period)
# at riskX 100
I =(365/w) Nrecent
Nseronegative + ½ (365/w) Nrecent
X 100
Survey size = 1000
HIV-seropositive = 100 (10%)
Recent on incidence assay = 10
RITA duration = 170 days
2.15 x 10Incidence = x 100 = 2.33% per year
900 + 21.5
HIV Incidence and RITA:Cross-Sectional Surveys
The Ideal Assay for Recent Infection• Describes a distinct “detection window” of relatively
uniform duration• Is universally positive in recent infection and negative
later in infection (or vice versa)• Is unaffected by:
– virus subtype– mode of transmission– therapy– OI and AIDS– Age, sex, race
• Has a relatively long window, all else being equal
0
0.5
1
1.5
2
2.5
3
3.5
0 200 400 600 800 1000 1200
RITA and Misclassification
Days
SOD
Mean
Mis-Classified
Long-standing
Misclassified Recent
170 days
Cutoff
-BED competitive capture EIA-Indirectly measures HIV-IgG as a proportionof total IgG
Amsterdam Cohort/ subtype B
0
0.5
1
1.5
2
2.5
3
3.5
4
0 200 400 600 800 1000
Days Since Seroconversion
OD
-n
127 days
Zimbabwe Seroconverters/C
0
1
2
3
4
0 200 400 600 800 1000
Days since SC
OD
-n
ENARP Serovonverters/C
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
0 200 400 600 800 1000
Days since SC
OD
-n
Kenya, A and D
0
0.5
1
1.5
2
2.5
3
3.5
4
0 200 400 600 800 1000
Days Since Seroconversion
OD
-n
181 days
167 days 171 days
Distribution of Window Periods for BED
from 768 seroconverters
-1
0
1
2
3
4
5
6
7
0 20 40 60 80 100
120
140
160
180
200
220
240
260
280
300
320
340
360
380
400
420
440
460
480
510
530
550
580
700
BED Window Period (Days)
Perc
enta
ge
Observed Estimates
Mean ω=197 days
5.9% > 2x mean ω to reach cutoff
Challenges to Using Antibody Maturation to Identify Recent Infection
• Variable immune response among individuals– Antibody response related to viral level
• Variability by HIV-1 subtypes• False-recent status (long-term specificity)
– Elite controllers (low viral levels)• Accumulate in population
– ART use (low viral levels)– Advanced HIV disease (AIDS)
2008
Proposed determination and use of a factor epsilon ( ε ) to correct for misclassification of long-standing infections as recent.
HIV Incidence Assays• “Detuned” assays
– Abbott 3A11 - unavailable– bioMérieux Vironostika HIV-1 – Avioq– Ortho Vitros ECi
• BED-Capture EIA (Calypte; Trinity)
• Avidity assays– Run on Abbott AxSYM– Bio-Rad – Run on Ortho Vitros analyzer
• IDE-V3 assay• IgG3 anti-HIV• Inno-LIA HIV adaptation
September 2008
What Needs to be Done• WHO Technical WG on HIV Incidence Assays
– www.who.int/diagnostics_laboratory/links/hiv_incidence_assay
• Guidance on assay use• Solidify consensus on mathematical issues• Define the assay development pathway• Define and assemble specimens for assays
calibration and validation• Engage industry on assay development
Assay Calibration and Validation• Establish “RITA Interval” and “False Recent Rate”• Requires large numbers of well-characterized
seroconversion panels and FRR panels– Various populations and sub-populations
• Geographic, transmission modes, etc.
– Various HIV-1 subtypes– Early and long-standing infections– Co-infections (TB, malaria)
• Such specimens aren’t readily available in sufficient volume in a central location
WHO HIV Technical Working Group on HIV Incidence
Guidance DocumentWhen and how to use assays for recent infection to estimate HIV incidence at a population level
Prepared on behalf of the World Health Organization Technical Working Group on HIV Incidence Assays With support of a grant from the Bill & Melinda Gates Foundation
Steps involved in applying RITA to estimate HIV incidence
Step 1: Identify study population
• The study population for which incidence will be calculated should be clearly defined.• The sampling frame that provides the subset for incidence testing should also be well defined.• Refer to Chapter 4 for guidance on this step.
Step 2: Identify HIV positive individuals
• Identify HIV positive individuals by testing the study population or the sampled subgroup for anti-HIV antibody or HIV RNA or DNA.
• Choice of tests and testing strategy should be based on local guidelines and consistent with WHO guidelines for HIV testing.
Step 3: Apply the Recent Infection Testing Algorithm (RITA)
• The RITA is applied to specimens of HIV infected individuals. The RITAs which may be utilised are:• RITA based on a single assay for recent infection • RITA based on a laboratory assay for recent infection combined with other clinical information
• Mean RITA duration must be known and false recent rate (FRR) must be calculated.• Refer to Chapter 6 for guidance on choice of RITA and the Appendix for the calculation of FRR.
Step 4: Analysis of resulting data
• HIV incidence is calculated using the counts from the RITA, the FRR and mean RITA duration.• Refer to Chapter 7 for guidance on estimating HIV incidence.
Application of a RITA based on a laboratory assay for recent infection and additional clinical information
HIV infected individuals
Assay for recent
infection*
Recent infection
Non-recent infection
Sample from recently HIV infected subject
CD4 count
Negative for AIDS illness
CD4 <200 cells/mm3
Negative
Non-recent infection
* Known mean RITA duration
Non-recent infection
PositiveNon recent infection
Anti-retroviral
testing
CD4 ≥200 cells/mm3
AIDS illness
Diagnosis of AIDS illness
Non-recent infection
History of HIV
infection
HIV diagnosis <1 year prior
Non-recent infection
Diagnosis ≤1 year prior or no previous HIV diagnosis
Performance in Clade B of BED + Avidity Testing Algorithm
Estimated Window Period using HIVNET & Vaxgen004 (154 subjects , median 4 samples / subject)
Avidity & BED cutoff Values
Mean Recency Period
(95% Conf Limits)
30% , 0.6 117.3 (93.1, 141.4)
40%, 0.8 143.0 (117.8, 168.2)
40%, 1.0 163.6 (138.0, 189.1)
50%, 0.8 147.8 (122.0, 173.5)
80%, 1.0 203.8 (170.0, 237.6)
Misclassification Rate in Known Chronically Infected Individuals (2 to 8 yrs post infection)Avidity & BED cutoff Values
MACS N= 341 individuals
ALIVEN= 284 individuals
30% , 0.6 0.58% (2/341)
0.35% (1/284)
40%, 0.8 0.74% (3/341)
0.35% (1/284)
40%, 1.0 0.74% (3/341)
0.35% (1/284)
50%, 0.8 2.35% (8/341)
0.35% (1/284)
80%, 1.0 3.81% (13/341)
3.52% (10/284)
Incidence Comparison at Johns Hopkins Emergency Department 2001 & 2007
Clade B epidemicUse BED 0.8 & Avidity 40%Assume 143 day window periodMisclassification rate of 0.7%
Survey year2001 2007
HIV negative 1366 4154HIV positive 183 321Recent positives 8 4
Incidence Estimates 1.26% 0.11%
One sided P value for difference = 0.0008
X
Acknowlegements• Oliver Laeyendecker• Sue Eshleman• Bharat Parekh• Bernie Branson• Andrea Kim• Connie Sexton• Mary Lynn Baniecki• Karine Dube• Megan Averill• Renee Ridzon• Bill Rodriguez• Christine Rousseau
• Tim Mastro• Chris Pilcher• John Kaldor• Txema Garcia-Calleja• Gaby Vercauteren• Thomas Rehle• Alex Welte• Tom McWalter• Tim Hallett• Michelle Owen• Joanne Micallef• Gary Murphy
